ComfyUI-rogala

Custom node pack for ComfyUI.

Installation

cd ComfyUI/custom_nodes
git clone https://github.com/rogala/ComfyUI-rogala

Restart ComfyUI. All nodes appear under the rogala menu.

Project structure

ComfyUI-rogala/
├── __init__.py          # Entry point — registers all nodes
├── pyproject.toml       # Package metadata
├── nodes/               # One .py file per node (or logical group)
│   └── _template.py     # Copy this to create a new node
├── js/                  # Frontend extensions loaded by ComfyUI
├── config/              # JSON configuration files
│   └── categories.json  # Category definitions
└── web/                 # Static web assets (CSS, icons) if needed

---

Aligned Text Overlay Images

Renders a multi-line text block onto an image at a chosen corner before saving. Supports %NodeTitle.param% template tags resolved from the active ComfyUI prompt.

---

Inputs

Pin	Default	Description
image	—	Input image tensor.
text_template	see default	Template string with optional %Node.param% tags.
vertical	bottom	Vertical anchor: top or bottom.
horizontal	right	Horizontal anchor: left or right.
font_size	16	Font size in points (10–50).
text_color	white	Text colour.
bg_color	black	Background colour (none = transparent).
bg_opacity	150	Background opacity (50–250).
external_text	—	Optional string appended after the resolved template.

Outputs

Pin	Type	Description
image	IMAGE	Outputs an image with a text block superimposed on top.

Example

Default template pulls values directly from a KSampler node:

seed: %KSampler.seed% | steps: %KSampler.steps%
cfg: %KSampler.cfg% | %KSampler.sampler_name% | %KSampler.scheduler%

NodeTitle must match the title shown on the node in the graph. Numeric sampler / scheduler indices are decoded to names automatically. Connect external_text from SamplerSchedulerIterator to append the current "sampler | scheduler" pair to the overlay.

---

Aligned Text Overlay Video

Renders a multi-line text block onto every frame of a video tensor. Supports %NodeTitle.param% template tags resolved from the active ComfyUI prompt.

---

Inputs

Pin	Default	Description
images	—	Video tensor (B, H, W, C).
text_template	see default	Template string with optional %NodeTitle.param% tags.
vertical	bottom	Vertical anchor: top or bottom.
horizontal	right	Horizontal anchor: left or right.
font_size	16	Font size in points (10–50).
text_color	white	Text colour.
bg_color	black	Background colour (none = transparent).
bg_opacity	150	Background opacity (50–250).
first_frame_only	false	Apply overlay only to the first frame (fast preview).
external_text	—	Optional string appended after the resolved template.

Outputs

Pin	Type	Description
images	IMAGE	Outputs a video tensor with text overlay applied.

Example

Connect between VAE Decode and video output:

VAE Decode -> AlignedTextOverlayVideo -> VHS Video Combine

Default template pulls values directly from a KSampler node:

seed: %KSampler.seed% | steps: %KSampler.steps% cfg: %KSampler.cfg% | %KSampler.sampler_name% | %KSampler.scheduler%

If you have multiple samplers, rename them in the graph (Right Click → Title) and reference explicitly:

steps: %Sampler_1.steps%

NodeTitle must match the title shown on the node in the graph. Numeric sampler / scheduler indices are decoded to names automatically.

Connect external_text from SamplerSchedulerIterator to append the current "sampler | scheduler" pair to the overlay.

Enable first_frame_only for fast preview (applies overlay only to frame 0).

---

Sampler Scheduler Iterator

Iterates over sampler x scheduler combinations one pair per execution. Outputs each pair to connected nodes (e.g. KSampler, Aligned Text Overlay Images). Node title updates automatically: Iterator: Step 3 / 12. Queue stops automatically after the last combination.

Note: Press Refresh before first use. This reads all installed samplers and schedulers from ComfyUI and writes the reference file. The counter resets to 0.

---

Inputs

This node has no inputs. Configure combinations in .\ComfyUI\custom_nodes\ComfyUI-rogala\config\sampler_scheduler_user.json.

Outputs

Pin	Type	Description
sampler_name	SAMPLER	Current sampler — connect to KSampler.
scheduler	SCHEDULER	Current scheduler — connect to KSampler.
external_text	STRING	`"sampler

Example

Edit config/sampler_scheduler_user.json to select combinations:

{
    "samplers": ["euler", "dpmpp_2m", "dpmpp_3m_sde"],
    "schedulers": ["karras", "simple"]
}

Total = samplers x schedulers (here: 3 x 2 = 6 runs). Copy names from config/sampler_scheduler.json (generated by Refresh). Unknown names are silently ignored.

---

LTX Resolution Selector

Selects the correct input resolution and frame count for LTX Video models. Supports Dev mode and Distilled upscale modes (x1.5 and x2).

---

Inputs

Pin	Default	Description
mode	Dev (no upscale)	Render mode — see table below.
dev_target	960x544	Target resolution for Dev mode (12 options).
upscale_target	1920x1080	Target resolution for Distilled modes (6 options).
fps	24.0	Frames per second (1–120). Passed through to output.
duration_sec	5	Clip duration in seconds (1–300).

mode options:

Value	Description
Dev (no upscale)	Standard LTX Dev sizes, rounded UP to nearest multiple of 32.
x1.5 Distilled	Returns input size needed to reach the target after x1.5 upscale.
x2 Distilled	Returns input size needed to reach the target after x2 upscale.

Outputs

Pin	Type	Description
width	INT	Input width in pixels (multiple of 32).
height	INT	Input height in pixels (multiple of 32).
length	INT	Frame count: 1 + 8 x round(fps x sec / 8).
fps	FLOAT	Passthrough of the fps input.

Example

To render a 5-second clip at 1920x1080 using x2 Distilled:

• mode = x2 Distilled
• upscale_target = 1920x1080 (Landscape)
• fps = 24, duration_sec = 5
• Node outputs: width=960, height=544, length=121, fps=24

---

Script for Aligning Nodes

A lightweight frontend extension that adds a persistent toolbar at the bottom of the canvas for aligning, distributing, and resizing selected nodes.

Features: align left / right / top / bottom · distribute horizontally and vertically (gap-aware, not center-based) · match width to widest node (aligns to leftmost) · deselect all

The toolbar auto-scales based on screen resolution (1080p / 2K / 4K) and includes manual size control (5 levels, −2 to +2) so the buttons stay comfortable at any display density. Localized tooltips (EN, UK-UA, DE, FR, ES, IT, PT-BR, ZH, JA)

Independent implementation inspired by KayTool (kk8bit) and ComfyUI-NodeAligner (Tenney95). No shared code — similar idea, different approach.